xlsx
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION] (MEDIUM): The script executes system-level commands using
subprocess.runto invoke LibreOffice (soffice) and timeout utilities (timeout/gtimeout). It also programmatically creates directories and writes files to the user's application configuration path. - [DYNAMIC_EXECUTION] (MEDIUM): The function
setup_libreoffice_macrogenerates a LibreOffice Basic macro (Module1.xba) as a string and writes it to the filesystem at runtime to enable theRecalculateAndSavefunctionality. - [PERSISTENCE] (MEDIUM): The skill modifies the user's persistent application environment by installing a macro into the LibreOffice
Standardlibrary. This script remains in the user's configuration directory after the skill execution completes. - [INDIRECT_PROMPT_INJECTION] (LOW): The script ingests external Excel files and iterates through all cells using
openpyxl. This represents a data ingestion surface where malicious content in a workbook could theoretically influence the agent's logic if the agent processes the resulting error summaries.
Audit Metadata